2026-03-10: Announcement: no advising sessions in week 4.
2026-03-16: Announcement: the advising session in week 5 will be held on-line, 2026-03-17T18:00: meet.google.com/zvo-yhcm-tdn
2026-03-23: Announcement: the advising session this week will also be on-line.
2026-03-24T18:00: meet.google.com/jut-xfyw-ith
2026-03-30: This week, as well, the advising meeting will be on-line.
2026-03-31T18:00: meet.google.com/bxu-dijk-rqx

Thesis advising slots (format: already_taken / total):
BSc. Year 2: 8 / 10.
BSc. Year 3: 8 / 8.
MSc.: 3 / 5

IMPORTANT: 2026-02-22: For students doing their Bachelor's Thesis with me, graduating in 2027: please send me an e-mail telling me your name and the subject we agreed to work on. Until the 1st of March 2026.
My advising interests are, in order of interest and expertise:
Genetic Algorithms
Artificial Neural Networks
general Machine Learning
Computer Vision
Numerical Physical (Mechanical / Electronic circuitry) Simulation
Astronomy and Astrophysics (co-advising)

My main requirement is to aim for an interesting, difficult thesis. I would rather receive ideas than be asked for thesis ideas, though these can arise through later discussion.
I do require a long time of working together; at a minimum, 1 semester, although I prefer 2 or 3 semesters of working on a thesis.
I do not add time deadlines for thesis delivery; meaning, in the unfortunate event of being unable to finish the thesis on time, I generally accept working together for 1 or 2 semesters more.

Advising meetings will be held according to my schedule, mixed on-line and on-site. Between my page (check my main tab here) and the schedule, you should have most information to contact me. If you have questions, please send me an e-mail with a subject containing the [thesis] tag.
If you're unable to make a meeting, consider creating a progress update as a .pdf or video presentation (it's also great to learn and practice how to explain and present clearly).

Template (on Overleaf) for a Bachelor's Thesis created by a student, in 2018.

Examples of theses I've advised:

• A GA-evolved config-based AI agent for playing the Spring RTS (BA) game. Able to defeat the standard Shard AI in 70% of matches. Link provided and maintained by the author.
• A Genetic Algorithm that uses a physics simulator to evolve parabolic trajectories, avoiding moving obstacles (dynamic optimum).
• A network traffic visualiser that classifies the traffic using Decision Trees, allowing for optional user ranking of features. (Used in the Industry, as an overview traffic viewer, to spot DDoS attacks and the like.).
• An ANN approach that shows a microphone can pick up distinguishable sounds for keypresses, lowering password brute force attacks to ~3^length (outside of a lab, in a typical working environment).
• A raytracing renderer, working as a Maya plug-in or stand-alone, multithreaded, focused on reflection quality and speed.
• A biased raytracer capable of greatly accelerating real-time raytracing performance, at the cost of visual accuracy.
• A cascade ensemble of Artificial Neural Networks, allowing remote-control with a simple laptop camera and hand gestures. Quick enough to play reflex games.
• A comparison between Q-Learning, NEAT and other methods for simple simulated racecars to find the optimal trajectory in a racing circuit. The race-circuit evolved through co-evolution.
• A physics simulator and ANN framework allowing various skeletal models to learn to walk, starting from only a reward function.
• A comparison of GAN and Autoencoder performance for colouring black-and-white portrait and nature landscape pictures.
• A quick Autoencoder-based solution for detecting and highlighting pulmonary disease on X-rays.
• A GAN-based solution for playing and generating levels for a 2D platformer (Mario-like).
• A thorough statistical analysis on investing in PUBG competitive players.
• Mixing evolutionary algorithms and backpropagation to optimise the structure, weights, and greatly lower the training time of Artificial Neural Networks. (Zmuschi, Shanti, and Eugen Nicolae Croitoru. "Optimising Artificial Neural Network topologies using Genetic Algorithms with very small populations." 2023 25th International Symposium on Symbolic and Numeric Algorithms for Scientific Computing (SYNASC). IEEE, 2023.)
• Accelerate the speed of Genetic Algorithms on the "Evolution of Mona Lisa" problem. (Vlad, Theodor-Alexandru, and Eugen Nicolae Croitoru. "Accelerating heuristic convergence on the" Evolution of Mona Lisa" problem by including image-centric mutation operators." 2022 24th International Symposium on Symbolic and Numeric Algorithms for Scientific Computing (SYNASC). IEEE, 2022.)
• Evolving 3D structures and "creatively" approximate hand-drawn targets. (Bucnaru, Raluca-Ioana, and Eugen Nicolae Croitoru. "Aesthetic Evolution of Target 3D Volumes in Minecraft." 2023 25th International Symposium on Symbolic and Numeric Algorithms for Scientific Computing (SYNASC). IEEE, 2023.)